AIAI Seminar-3 June-Talk by Guillermo Romero Moreno

Speaker: Guillermo Romero Moreno 

Title: A Methodological Overview of Disease Network Inference from Electronic Health Records: The Quest of Capturing Associations between binary variables

Abstract: The prevalence of multimorbidity, or the accummulation of two or more long-term conditions in the same patient, has been increasing due to ageing populations. There is a need to better understand and manage the interaction between the health conditions, as current medical healthcare and research has a narrow specialist approach. An alternative is to use system-wide exploratory approaches based on big data that are able to uncover important patterns and connections between the conditions. Some of these approaches attempt to detect the possible associations between the conditions given in the data and assemble them into networks that serve as 'maps' of the multimorbidity landscape and that can serve to develop further hypotheses. In this talk, I will provide a taxonomy of the approaches for building phenotypic disease networks used in the multimorbidity literature. As the underlying problem is the general one of finding associations between binary variables (an instance of the inverse Ising problem) the approaches can be linked to many other application domains, such as psychonometrics, genetics, or the social sciences.